Decision Feedback Equalizer Crashes Matlab with InputSamplesPerSymbol option

Question

Arda Karaca 2021-5-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/820550-decision-feedback-equalizer-crashes-matlab-with-inputsamplespersymbol-option

评论： Arda Karaca 2021-5-16

I make use of Decision Feedback Equalizer from Communications Toolbox with PAM-2 (OOK) signal. In my setup each symbol has 10 samples and I used "InputSamplesPerSymbol" parameter in the equalizer initializaiton.

DFE compiles succesfully however when I call the DFE with data, MATLAB crashes without any warnings/errors/output.

I have tried this on another setup as well and the result is no different. What can be a potential cause and what would be a possible workaround or a solution?

%% Generate Samples
numTrainSymbols = 200;
numDataSymbols = 1800;
numSamples = 10;
trainingSymbols = real(pammod(randi([0 1],numTrainSymbols,1),2));
dataSymbols = real(pammod(randi([0 1],numDataSymbols,1),2));
stream = repelem([trainingSymbols; dataSymbols], numSamples);
SNR = 20;
ch_out = awgn(stream, 20);
%% Input Data
dfeq = comm.DecisionFeedbackEqualizer('Algorithm','LMS', ...
                                        'NumForwardTaps',5,...
                                        'NumFeedbackTaps',4,...
                                        'ReferenceTap',5,...
                                        "Constellation", [-1 1], ... 
                                        'StepSize',0.001, ...
                                        "InputSamplesPerSymbol", numSamples);
                                        
[y, err, weights] = dfeq(ch_out, trainingSymbols);

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Sindhu Karri 2021-5-13

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/820550-decision-feedback-equalizer-crashes-matlab-with-inputsamplespersymbol-option#answer_698703

在 MATLAB Online 中打开

Hii,

I tried to run the code in MATLAB R2021a and it gave the following error message.

numTrainSymbols = 200;
numDataSymbols = 1800;
numSamples = 10;
trainingSymbols = real(pammod(randi([0 1],numTrainSymbols,1),2));
dataSymbols = real(pammod(randi([0 1],numDataSymbols,1),2));
stream = repelem([trainingSymbols; dataSymbols], numSamples);
SNR = 20;
ch_out = awgn(stream, 20);
%% Input Data
dfeq = comm.DecisionFeedbackEqualizer('Algorithm','LMS', ...
                                        'NumForwardTaps',5,...
                                        'NumFeedbackTaps',4,...
                                        'ReferenceTap',5,...
                                        "Constellation", [-1 1], ... 
                                        'StepSize',0.001, ...
                                        "InputSamplesPerSymbol", numSamples);
                                        
[y, err, weights] = dfeq(ch_out, trainingSymbols);
Error using comm.DecisionFeedbackEqualizer/validatePropertiesImpl (line 166)
NumForwardTaps must be greater than or equal to InputSamplesPerSymbol.

Changing the value of 'NumForwardTaps' property to some value like 15 which is greater that numSamples helps to resolve the issue.

When I executed the same code in R2020a MATLAB crashes.So,I recommend you to update the MATLAB since this issue fixed in later releases.

Refer to below link to update the MATLAB

https://www.mathworks.com/help/install/ug/upgrade-matlab-release.html

Hope this helps!!

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Arda Karaca 2021-5-16

Exactly. I later contacted MATLAB technical support. They let me know that 2020a does not check for inputs properly and this issue is later fixed in the new version. Thank you for your reply!

请先登录，再进行评论。